Latency & Jitter Meter
Latency and jitter, measured properly — the numbers a speed test hides.
Run the test
Round-trip time of an HTTP request to the nearest Cloudflare edge, sampled repeatedly. Not an ICMP ping.
How this check works
Twenty-five timed requests to the nearest edge, reported as a distribution rather than an average. The median is what a well-behaved connection does; the 95th percentile is how bad the slow tail gets; and the jitter — the mean change between consecutive samples — is the figure that decides whether a call sounds natural. That last one is the reason this page exists separately from a speed test: a connection can average 30 ms and still deliver individual packets 200 ms late, which is exactly what a jitter buffer struggles with and exactly what an average conceals.
What the results mean
- Median
- The typical round trip. Under 60 ms is excellent for a domestic connection, and above 150 ms starts to be felt in conversation.
- Jitter
- The mean change between consecutive samples, not the spread. An alternating connection and a drifting one score alike on deviation and feel completely different.
- 95th percentile
- The slow tail. A good median with a poor 95th is a connection that stutters rather than one that is slow.
- Loss
- Requests that did not complete. Any loss at all is worth investigating; it damages real-time traffic more than latency does.
Common problems and fixes
- My ping in a game is much higher than this
- Expected. This measures to a nearby edge; a game server is further away and behind more hops. The useful comparison is between measurements over time, not against a game.
- The numbers are worse on Wi-Fi
- Very common, and worth quantifying before doing anything else. Wi-Fi adds jitter through interference and airtime contention, and a cable removes it.
- It looks fine but calls are still bad
- Try it while something else is downloading. Bufferbloat only appears under load, and the bufferbloat test is built for exactly that case.
Frequently asked questions
What is a good ping?
Under 60 ms to a nearby server is excellent for a domestic connection, 60 to 150 is fine for everything except competitive gaming, and above 150 starts to be noticeable in conversation. The more useful number is jitter: a steady 120 ms is a perfectly good call, and 40 ms that swings by 60 is not.
Why is this different from the ping command?
A browser cannot send the low-level probes that command uses, so this times an HTTP request instead — which includes the server's own handling time and a little protocol overhead. It reads a few milliseconds above a command-line ping to the same place, and it moves in the same way when something changes, which is what makes it useful.
What causes high jitter?
Wi-Fi interference and airtime contention first, then anything sharing the connection, then congestion on the provider's network. In that order, and the first two are the ones you can do something about this afternoon.